Dark Arts 101: Be rigorous, on average

I’m reading George Steiner’s 1989 book on literary theory, Real Presences. Steiner is a literary theorist who achieved the trifecta of having appointments at Oxford, Cambridge, and Harvard. His book demonstrates an important Dark Arts method of argument.

So far, Steiner’s argument appears to be:

  1. Human language is an undecidable symbol-system.

  2. Every sentence therefore carries with it an infinite amount of meaning, the accumulation of all connotations, contexts, and historical associations invoked, and invoked by those invocations, etc. Alternately, every sentence contains no meaning at all, since none of those words can refer to things in the world.

  3. The meaning of a sentence, therefore, is not finite or analyzable, but transcendent.

  4. The transcendent is the search for God.

  5. Therefore, all good literature is a search for God.

The critics quoted on the back of the book, and its reviews on Amazon, praise Steiner’s rigor and learning. It is impressive. Within a single paragraph he may show the relationship between Homer, 12th-century theological works, Racine, Shakespeare, and Schoenberg. And his care and precision with words is exemplary; I have the impression, even when he speaks of meaning in music or other qualia-laden subjects, that I know exactly what he means.

He was intelligent enough to trace the problems he was grappling with out past the edges of his domain of expertise. The key points of his argument lie not in literary theory, but in information theory, physics, artificial intelligence, computability theory, linguistics, and transfinite math.

Unfortunately, he knows almost nothing about any of those fields, and his language is precise enough to be wrong, which he is when he speaks on any of those subjects. How did he get away with it?

Answer: He took a two-page argument about things he knew little about, spread it across 200 pages, and filled the gaps with tangential statements of impressive rigor and thoroughness on things he was expert in.

For example, the first chapter discusses, with perhaps a hundred references, his opinion that the best art criticism is art made in response to art, his theory that good art is always derived from earlier art, and observations on the etymology of words; but most especially his consternation at the hundreds of thousands of articles, books, and talks on literature produced yearly by people who are not professors at Cambridge, Oxford, or Harvard. Then on page 36, he says,

The positing of an opinion about a painter, poet or composer is not a falsifiable proceeding.

I think this is the only sentence in the chapter that is a crucial part of his argument. But instead of engaging with the body of literature on what falsifiable means, whether human language is falsifiable in general (is “Ben is tall” falsifiable?), and what falsifiability has to do with the communication of information, Steiner lowlights this sentence with its syntactic simplicity, and nearly buries it in complex, learned sentences about Dante, Mozart, and hermeneutics. We dive back into litspeak, to emerge again into his argument on page 61:

All elucidation and criticism of literature, music and the arts must operate within the undecidability of unbounded sign-systems… Talk can be neither verified nor falsified in any rigorous sense.

Here one should ask: If human language is recursively enumerable, why don’t people understand sentences generated by context-free grammars for English when they go past one level of recursion (“The mouse the cat the dog chased chased squeaked”)? And doesn’t “undecidable” mean “there exists at least one undecidable sentence” rather than “all sentences are undecidable”?

But one does not; one goes on to Steiner’s opinions of Tolstoy’s opinions of King Lear. It will be nearly another 20 pages before we hit the next key point in his argument, which relies on his not knowing that one can compute the sum of some infinite series. The bulk of the first 90 pages [1] is impressive displays of learning which fill in the vast spaces between the points (almost literally) of his argument.

Unless a reader pays close enough attention to catch these brief ventures outside Steiner’s areas of expertise, he will come to the end of the book with (A) a summary of Steiner’s argument, and (B) the strong impression that the statements in the book were learned and rigorous. And thus, the argument carries.

ADDED: After looking at the long section on the impossibility of meaning that begins around page 90, it seems Steiner is not trying to argue points 1 and 2 at all when he refers to them in chapters 1 and 2. He is merely foreshadowing. On p. 102-103 we reach the heart of his defense of points 1 and 2, which is to say “Wittgenstein said so.” I’m afraid that, if I finish the book, I might find a better summary of the method here to be that the key points of his argument are defended only by appeals to authority.


[1] This pattern breaks down around page 90, where Steiner begins a long spiral into his central thesis.